Printer Friendly

Servicing under a microscope.


Isolating the key portfolio traits linked to higher servicing costs requires careful scientific probing. KPMG Peat Marwick performed such analysis in its 1990 servicing performance study.

THE FIFTH YEAR OF KPMG Peat Marwick's Mortgage Servicing Performance Study (MSPS) yielded results that reflect many of the common findings from the prior four years' studies. One example of this consistency of results is that productivity proved once again to be one of the strongest driving forces behind low-cost servicing performance.

With a new year, new participants and renewed vigor, we further refined the quality of data collected and the analysis performed in the 1990 study. In doing so, we were able to approach new frontiers in cost and performance relationships within mortgage servicing institutions.

Using the Statistical Package for the Social Sciences (SPSS) as the analytical tool, we began to "peel" apart the mortgage servicing operation into finer and finer segments; trying to identify new relationships among the wealth of data gathered in this year's study. This analytic process brought us closer to the core factors influencing the performance of the servicing operation.

One of the main contributors that differentiated low-, mid- and high-cost performance was portfolio composition. Portfolio composition is a broad term that may signify any number of factors, but for simplicity's sake, we will highlight the following five primary factors that distinguished low- and high-cost operations.

The total portfolio for this year's study participants consisted of 3 million loans, of which 26 percent were serviced for GNMA, 20 percent for Fannie Mae, 16 percent for Freddie Mac, 13 percent for private investors and 25 percent for parent, owned and other.

GNMA and core cost

Pervasive within the industry is the theory that large GNMA servicing operations take on "factory-like" characteristics. As the loans are fully escrowed, all loans are "handled the same" for tax and insurance purposes. The rigors of the FHA-claim process necessitates exception-proof processing of defaults. The impact of VA no-bid losses places added demands on the management of delinquent loans, fosters further geographical sensitivities and so forth.

Indeed, we have found that these larger GNMA servicers are hardly factory-like at all. Typically, the lower cost servicers perpetuate a philosophy of "there ought to be a better way." Significant emphasis is placed on automation to handle the vast majority of loans and to ensure adequate resources to deal with the true exceptions. These servicers focus on loss mitigation and have done so for years. Also, these servicers tend to manage by exception. As the percentage of exceptions increases, so do their costs. These servicers control costs by controlling the number of exceptions, for example and demographics, by profiling borrower characteristics to ensure timely collection efforts.

Consistent with 1989, the lower cost servicers had a higher percentage of servicing for GNMA (42 percent) than any other investor type. As the percentage of GNMA servicing increased, there was moderate downward pressure on the cost of servicing.

More interesting was the very strong, negative relationship between the percentage of loans serviced for GNMA and the cost of the data processing and customer-inquiry functional areas. These two functional areas, which represented nearly 25 percent of overall core cost in this year's study, declined noticeably as the percentage of GNMA servicing increased. This finding suggests that these two functional areas benefit greatly from minimizing exceptions associated with servicing the FHA and VA loan products

For example, the data processing associated with FHA and Va loans is consistent for all FHA and VA loans, with minimal variations. In reality, some servicing systems may have been developed with the FHA or VA product in mind. (Conversely, a conventional product may vary by term, coupon, index, margin and the like.) Unlike data processing, customer inquiry is a highly variable function. However, the customer-inquiry function may also benefit from the familiarity and consistency of the FHA and VA loan products, which in turn makes it easier to respond to the customer's questions.

These characteristics, along with the fact that the large GNMA servicers tended to have a below-average number of ARM loans and owned loans, all contributed to the lower core cost for the larger GNMA servicers in the 1990 study.

Fannie Mae, Freddie Mac and private investors

Results from the 1990 MSPS study suggest that a shift in the percentage of Fannie Mae, Freddie Mac or private investor servicing had a slight, positive impact on the overall core cost. When plotted, there were some very weak relationships between the percentage of loans serviced for Fannie Mae and Freddie Mac and core cost for the customer service functions; however, this upward pressure on cost reflected, in all likelihood, the cost of servicing a wide variety of loan products rather than the cost of servicing for either agency.

The lack of relationship among core costs and servicing for particular investors is in decided contrast to the results of the 1989 study. In that study, private investor loans behaved very much like GNMA loans with regard to core cost; placing downward pressure on the cost of servicing as the percentage of private-investor servicing increased. That relationship was weak, at best, in 1990. In addition, the 1989 study found upward pressure on the cost of servicing as the percentage of Freddie Mac servicing increased. That particular relationship was not found in 1990, reflecting perhaps a one-time impact on the cost of reporting changes implemented by Freddie Mac in 1989.

Owned loans and core cost

Consistent with findings in the 1989 study, owned loans represented the largest percentage of loans serviced for the higher cost servicers. The findings in the 1990 study, however, exhibited an even stronger positive, or upward, pressure on overall core costs from this type of servicing.

Just as interesting was the strong, positive relationship between an increase in the percentage of owned loans serviced and the cost of the functional areas of customer service, collections and foreclosure and ancillary functions. Analyzing these relationships in more detail revealed that the strongest positive relationship in the study occurred between the percentage of owned loans and the cost of the customer-inquiry and special-loans areas. Because customer inquiry is one of the higher cost areas, any upward pressure on this function must be carefully managed. The reasons why owned loans (other than loans bought out of pools for very specific reasons) should cost more to service than an agency or private-investor loan lacks a clear explanation.

Breakdown by loan type

Not surprisingly, analysis of the relationship between product type and core cost produced results similar to those of the agencies who bought the loans. When reviewing the relationships of the different loan types on an individual basis, the following was noted:

* As the percentage of FHA loans

serviced increased, there was

increased downward pressure on the

core cost. * As the percentage of VA loans

increased, there was downward

pressure on key core-cost functional

areas with the exception of

collections and foreclosure. An increasing

VA portfolio put moderate upward

pressure on the cost of collections,

foreclosure and REO. This finding

is supported by other

characteristics of the 1990 study portfolio,

including the fact that VA loans had

the highest default rate and the

highest loan-loss unit cost (caused

by no-bids and buydowns). * Conventional loans had mixed

results, being positive and negative

depending on the area. Overall, an

increasing percentage of

conventional loans placed strong, upward

pressure on the cost of customer

service and ancillary functions.

More specifically, both the data-processing

and customer-inquiry

functional costs grew as the

conventional portfolio grew.

As these functions

represent the

second- and fourth-highest

cost areas in the study, it follows

that higher cost servicers had a

higher percentage of conventional

loans. * Participants with large FHA/VA

servicing portfolios were expected to,

and did, have higher overall

delinquency rates than participants who

had large conventional portfolios.

However, when calculating an

individual delinquency rate for each

loan type, we found that the

FHA/VA servicers had lower

delinquency rates than large

conventional servicers for all three loan types

(FHA, VA and conventional). The

reason for the overall higher

delinquency rates among FHA/VA

servicers was due to FHA and VA loans

having inherently higher default

rates than conventional loans. This

fact led us to believe that the large

FHA/VA servicers expected higher

delinquency rates and have

developed effective collection techniques

to help mitigate the eventual default


Portfolio by year of origination

Yet another layer of analysis was performed on the 1990 study portfolio to discern the impact of the age of loans on the core cost of servicing. The results suggest that newer loans (one to two years old) might be less expensive to service than slightly older loans (three to five or more years old) based on inherent characteristics of the loan's age.

In the first years of the loan, certain functional areas, particularly customer service, tend to provide more support than other functional areas, such as the collections and foreclosure area. Customer-inquiry, tax and insurance functions are key resources called upon by the borrower in the earlier years.

In years three through five, the collections and foreclosure function plays a more active role in the servicing of the loans, as the loans have entered their peak delinquency and foreclosure period. These loans require significant servicing attention.

Many would suggest that a newer loan requires more, not less, interaction with the borrower than a three- to five-year-old loan, and hence, has higher costs. Others would suggest that a newer loan, properly set up, requires less borrower interaction, and hence has lower costs. Our findings suggest that a newer loan may be less costly than an aging one, because the loan is serviced predominately by a lower servicing-cost functional area--the customer-service functional cost is $26 per loan.

Conversely, an older loan has a greater likelihood of spending more servicing time in a more costly functional area--collections and foreclosure--which as a function averaged $39 per loan (excluding loan losses and indirect expenses). Hence our findings suggest that newer loans are less expensive to service than older loans, because newer loans are spending more time in less-expensive servicing areas.

The year of loan origination also supported the premise that newer loans might be less expensive to service than slightly older loans in the following ways:

* As the percentage of newer loans

(originated in 1989 and 1990)

increased, there was downward

pressure on overall core cost and core

costs in each functional area.

Stratifying the participants into lower,

mid- and higher cost servicers, the

lower cost servicers tended to have

a larger portion of their portfolio

originated in 1989 and 1990. The

higher proportion of newer loans

also represented the portfolio

growth characteristic among the

lower cost servicers in both the

1989 and 1990 MSPS Study. * Loans originated in 1987 and 1988

produced positive or higher cost

relationships. These loans were

entering the initial peak delinquency

phase, resulting in additional

servicing costs. The percentage of loans

originated in and prior to 1986

displayed no statistically significant

impact on the cost of servicing. * As expected, there was a slight

relationship between overall

delinquency and foreclosure rates, the year of

origination and core cost. The lower

cost servicers had a smaller

percentage of loans originated in the

1986-to-1988 period, loans that

would normally be considered in

the peak delinquency and

foreclosure time period. Appropriately,

overall delinquencies were slightly

lower for lower cost servicers

than higher cost servicers: 4.82

percent versus 5.10 percent for

delinquency; .91 percent versus

1.24 percent for foreclosure;

and .05 percent versus .20

percent for real-estate owned

(REO). These higher default

rates translated almost

directly into higher average core

costs in the collections and

foreclosure area, the

highest cost area in the study.

Geographical diversification and core cost

Throughout the study's history, we have questioned whether the geographical diversification of a servicing portfolio had an impact on the cost of servicing. In this year's study, overall core cost was relatively unaffected by servicing a portfolio of predominantly California, New York or even Florida loans. Cost differences were noted in certain functional areas, such as taxes, assumptions and pay-offs, but there was not a strong, overall relationship.

However, when looking at servicers whose portfolio concentration was limited to one or two states, some differences were noted. In general, it was found that the lower cost servicers had a greater dispersion of loans across the states; conversely, higher cost servicers had a higher concentration of their servicing portfolio in one state. For example, those participants whose portfolio concentration in their largest servicing state did not exceed 30 percent had a below-average core cost. Participants who concentrated 75 percent or more of their portfolio in one state had an above-average core cost. The findings suggest that the lower cost (and less concentrated) participants used more outsourcing of servicing functions than higher cost (and more concentrated) participants. The higher cost (and more concentrated) participants had 48 percent of their core costs in personnel (versus 38 percent of lower cost and less concentrated participants), the largest portion of overall core costs.

The 1990 KPMG study results consistently demonstrate that the geographic location of the servicing operation has a significant impact on core costs. Basically, having a servicing operation in a more remote location will help minimize labor, occupancy and other operating costs. The higher cost servicer spent $25 more per loan on personnel than the lower cost servicer. In addition, the higher cost servicer incurred $11 more per loan in other operating expenses, which includes occupancy, utilities and the like.

Form of organization

Another interesting piece of information that began to surface last year and was brought to the forefront in this year's study was the relationship between the form of organization--wholly owned subsidiary, bank or thrift--and both costs and portfolio characteristics. The average core cost for wholly owned subsidiaries of a bank, thrift or other institution was 8 percent below the overall average core cost for all of this year's study participants. Conversely, banks and thrifts, whose mortgage operations are an integrated part of the bank or thrift, had overall core costs 20 percent above the overall average.

When analyzing the allocation of resources for the two groups, the wholly owned subsidiaries allocated 39 percent of their resources to personnel, compared to 43 percent for the banks and thrifts, and 41 percent to operating expenses, compared to 35 percent for the banks and thrifts. This suggests the use of more outsourcing by the wholly owned subsidiaries, which is probably a function of, in part, the wider geographic dispersion of the loan portfolio.

This relationship did not hold true, however, for noncore costs, with the exception of overhead allocation, where the wholly owned subsidiaries' costs were half that of the banks and thrifts. The other noncore costs for the wholly owned subsidiaries, which include loan losses, interest on advances and interest on escrows, were 50 percent or more above that of the banks and thrifts. In aggregate, therefore, the average total servicing cost for wholly owned subsidiaries was 7 percent greater than that of the banks and thrifts.

Comparing overall delinquency, foreclosure and bankruptcy rates as a percentage of the overall portfolio between wholly owned subsidiaries and banks and thrifts, the banks and thrifts had lower overall ratios in all three categories. This may indicate a loan quality difference in the servicing portfolios because the banks and thrifts originated (retail) 72 percent of their current portfolio versus 49 percent retail origination for the wholly owned subsidiaries.

Other portfolio characteristics that differed among the two groups included investor makeup and geographic dispersion. The banks and thrifts averaged 35 percent owned loans and a 71 percent concentration of loans in one state compared to 33 percent GNMA and a 28 percent single-state concentration for the wholly owned subsidiaries. As mentioned earlier, these two characteristics in particular had a significant affect on overall core costs in this year's study.

The relationship between the various portfolio characteristics and the cost of servicing serve as interesting predictors of core cost overall and for several key functional areas. KPMG Peat Marwick spends many hours trying to build an understanding of lower versus higher cost servicing characteristics. Difficult to summarize, in any context, is the definition of desired return for these servicing operations. We are aware, for instance, of the form of ownership, targeted returns and other various issues that affect participants' performances in a given study year. Our thorough analysis only partially reflects the repetoire of strategies (i.e., growth, diversification) of companies and is a snapshot of where individual participants are within their implementation program. But after five years, we have developed an extensive collection of experiences from which characteristics of lower, mid- and higher-cost servicers are clearly emerging. What is clear is that lower costs do not always translate to a higher return, just as higher costs do not always translate to a lower return.

The bottom line for a study such as MSPS is to build a better understanding of the servicing challenges and the participants' responses, and to provide a timely forum for such an exchange. The Mortgage Servicing Performance Study Class of 1990 receives high marks for their efforts to both provide and exchange. All share one characteristic: clearly, they are finding better ways to service mortgage loans.

Linda C. Simmons is a principal in KPMG Peat Marwick's Washington, D.C. office and Michele M. Long is a senior manager in the Washington, D.C. office.
COPYRIGHT 1991 Mortgage Bankers Association of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:isolating key portfolio traits linked to higher servicing costs requires careful scientific probing
Author:Simmons, Linda C.; Long, Michele M.
Publication:Mortgage Banking
Article Type:Cover Story
Date:Sep 1, 1991
Previous Article:Boardroom view.
Next Article:Perfecting the prepayment hedge.

Related Articles
Muscular dystrophy gene cornered.
Imaging ionic tides and soft surfaces.
Making 'movies' of biological molecules.
Servicing's numbers game.
Get proactive.
Random Samples.
Horizontal Gene Transfer Happens.
Veeco Buys ThermoMicroscopes.
American Scientific Publishers (ASP; Stevenson Ranch, CA) has launched "Journal of Scanning Probe Microscopy," a new quarterly journal that covers...
Tool set improves performance.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters